FR3040156A1 - AUTOMATED BRAKE ACTUATOR CONTROL METHOD AND VEHICLE OR VEHICLE BRAKE USING THE SAME - Google Patents

Abstract

The invention relates to a method for controlling an actuator of a vehicle brake, comprising an iteration of the following steps: from a value of the desired clamping force, determination of a set value y * ( t0) targeted for an output quantity; applying a control variable of value u (t0) at a time t0; measuring a value found y (t0) of the output quantity produced within the actuator under the effect of said application, and as long as an adjustment condition is true, calculating a new value u (t1) for the control variable, at a new instant t1, so as to achieve a command without modeling the actuator Proportional-Integral correction type "Intelligent", called "modelless control". The invention also relates to a vehicle or vehicle brake implementing this method.

Description

The invention relates to a method of controlling a vehicle brake actuator, implemented by a digital computer onboard the vehicle. within said vehicle.

This control method comprises an iteration of the following steps: from a value of the desired clamping force, determination of a desired target value y * (tO) for an output quantity; applying a control variable of value u (t0) at a time t0; measuring a value found y (tO) of the output quantity produced within the actuator under the effect of said application, and as long as an adjustment condition is true, calculating a new value u (tl) for the control variable, at a new instant tl, so as to achieve a control without modeling the actuator by Proportional-Integral correction of "Intelligent" type, called "command without model".

It also relates to such a method applied to an automated parking brake. The invention also relates to a vehicle or vehicle brake implementing this method.

State of the art

To control the operation of a motorized brake in a vehicle, for example by an onboard computer, it is known to use a method of using a mathematical modeling of the behavior of the brake and the actuator according to the control parameters that it receives, for example electrical voltage and intensity. From the modeling of this behavior, we create a control law that will be used to control the actuator by sending the appropriate electrical parameters.

This law, however, must itself be adjusted and parameterized so that the actual behavior, in situation, meets the expectations of the designer, without requiring too much performance on the part of the components, for example engine power and mechanical strength of the components. parts, because of the costs and constraints associated with such a component requirement.

Currently, the estimate of the clamping force applied is based on a back-calculation using physical equations that describe the physical structure of the brake and the actuator, for example the gear ratio of the various components involved.

An object of the invention is to overcome all or part of the disadvantages of the prior art, and in particular to improve the reliability and accuracy of the behavior obtained in real-life situations, in particular under severe conditions and for significant environmental variations.

Presentation of the invention

This object is achieved with, according to a first aspect of the invention, a method for controlling an actuator of a vehicle brake, implemented by a digital computer embedded in said vehicle to produce an application by said brake of the vehicle. a determined clamping force, comprising an iteration of the following steps: from a value of the desired clamping force, determination of a desired target value y * (tO) for an output quantity, - application a control variable of value u (t0) at a time t0 at the input of the actuator, - measurement of a value found y (tO) of the output quantity, produced within the actuator under the effect of said application, and - as long as an adjustment condition is true, calculating a new value u (tl) for the command variable, at a new instant tl, and applying this new input value of the actuator, which value depends on: • the value u (tO) of the control variable, • the derivative ÿ (tO) of the output quantity, • the rate of change y * (tl) of the setpoint value, between tO and tl, • a factor proportional to the error of follow-up of the output quantity, • a factor proportional to the integral of the error of tracking of the output quantity, said new value u (tl) being able to be written in the following form:

where: • a, Kp, and Ki are predetermined coefficients, • e (t) = y (t) - y * (t) is the tracking error; so as to achieve a command without modeling the actuator Proportional-Integral correction type "Intelligent" called "modelless control". The invention thus proposes a method of controlling a motorized brake actuator, in particular but not necessarily to control the operation of an automated parking brake.

The adjustment condition can be varied. It is for example defined with respect to the tracking error. Preferably, the method continues as long as the absolute value of the tracking error decreases.

This is a model that could be described as empirical, in the sense that it does not require, or less, to model the behavior of the brake in a mathematically precise way.

This method uses data determined and developed by a form of learning on a prototype, with recording of the results found in association with the parameters applied to obtain these results. It is no longer necessary to know all the operating equations, or less necessary.

This type of control without mathematical model is possible in particular because it applies to a linear operation system for example which does not include a sudden change cam or components operating in all or nothing.

This method can be implemented without a "preferred" sensor, in "open loop". It can also be implemented with choice of a preferred sensor, for example with a Hall effect sensor on the speed, which is related to the position of the actuator.

The control mode of the electric motor is preferably in pulse width modulation ("Pulse Width Modulation"), in particular for a DC motor (DC), for example with a serial current control. It can also be done in vector control, for example for a brushless motor ("brushless").

This method allows a better robustness and reliability of the results obtained, in particular in real situations, and a better precision of the level of effort applied, with less influence of the environmental variations.

Various embodiments of the invention are provided, integrating, according to all of their possible combinations, the various optional features set forth herein.

In one embodiment, the setpoint y * is determined from a correspondence list comprising a plurality of intensity values each associated with a value of a clamping force produced by said intensity within said actuator.

Advantageously, the control method comprises, from an initial intensity value chosen as a function of a clamping force to be applied, an iteration of the following steps: comparison of said initial value with a database forming the list of correspondence and containing a series of intensity values associated with a clamping force obtained during a preliminary learning phase, for calculating a modified intensity value according to an adjustment coefficient, - sending a control command supplying the motor (for example by adjusting the voltage) based on this modified intensity value, according to at least one determined method of motor management, in particular in pulse width modulation or in vector control, minus one parameter of result within the actuator, - estimation of a gain found for said modified intensity value, - according to said observed gain, comparison with said base data to calculate a modified value of the adjustment coefficient.

Preferably, the control method measures a clamping force obtained after application of the control variable u from the setpoint y * and an update of the clamping force associated with said setpoint y * in the correspondence list.

According to a particularity, the setpoint y * is a current intensity setpoint to be obtained in the brake actuator.

According to one possibility, the value found is obtained by measuring the current intensity as generated in the brake actuator. This measurement is preferably direct and / or without intermediate size.

Preferably, the derivative ÿ of the observed value is determined in the form of an estimate y (t), where:

where T is a predetermined parameter.

According to one embodiment, the estimate is calculated by discretizing its integral, for example by the trapezoid method. Any other method can be used, including the rectangles method.

According to another aspect of the invention, there is provided a vehicle or vehicle brake, comprising at least one computer programmed to control one or more actuators according to a method according to the first aspect of the invention or one of its improvements.

DESCRIPTION OF THE FIGURES Other features and advantages of the invention will appear on reading the detailed description of implementations and non-limiting embodiments, with reference to the appended figures in which: FIG. 1 is a diagram illustrating a embodiment of a method 100 according to the invention, - Figures 2 to 6 are presentation boards that specify the application of the method to a brake actuator in general, - Figures 7 to 12 are boards of presentation that specify the application of the method to an automated parking brake actuator.

Description of the invention

These embodiments being in no way limiting, it will be possible in particular to make variants of the invention comprising only a selection of characteristics described hereinafter, as described or generalized, isolated from the other characteristics described, if this selection of characteristics is sufficient to confer a technical advantage or to differentiate the invention from the state of the art.

The method 100 according to the invention controls an actuator block 1, also called "System" in the figure, of a vehicle brake.

According to an embodiment of an implementation of the method 100, the method 100 is implemented by a digital computer embedded in the vehicle, in particular to produce an application by the brake of a determined clamping force.

The method 100 comprises an iteration of the following steps: from a value of the desired clamping force, a determination 102 of an intensity y * (t0), represented by the letter I * in FIG. for an output quantity, - an application 104 of a control voltage of value u (t0) at a time t0, - a measurement 106 of a value found y (tO), represented by the letter I in FIG. of the output quantity produced within the actuator 1 under the effect of said application, and - as long as an adjustment condition is true, calculating a new value u (tl) for the magnitude of control, at a new instant tl, which value depends on: • the value u (tO) of the control variable, • the derivative ÿ (tO) of the output quantity, • the rate of change y * (tl) of the setpoint, between tO and tl, • a factor proportional to the error of follow-up of the output quantity, • a factor proportional to the integral of d e the tracking error of the output quantity.

In the example represented in FIG. 1, at the index k, that is to say at a time t1, the calculation of the new value is determined by the implementation of three functional blocks, respectively a correction block 2 intelligent proportional integrator type, a voltage generator block 3 and an estimator block 4, in addition to the actuator 1.

The actuator block 1, also called "System" in the figure, comprises, according to the representation of Figure 1, an input U (not shown) which is a voltage, connected to a voltage input noted "voltage [V]" on the figure, and has an output I, which is an intensity, connected to an output marked "intensity [A]" in the figure.

The correction block 2, also called "i-PI" in the figure, comprises three inputs, respectively denoted I *, G (k), and I, and an output denoted u (k). The input I * is received from the step 102 of the setpoint determination. The input G (k) is connected to an output G (k) of the estimator block 3.

The output u (k) of the correction block 2 is equal to a summation of: a signal coming from a full proportional type corrector on an input control I *, which can be written as:

with Kp, and Kj which are predetermined coefficients of gain of the integral proportional type corrector, and e (t) = I (k) - V (k) which represents the tracking error; and a signal originating from the multiplication by a gain equal to the inverse of a coefficient a, of a signal of difference between that resulting from a rate of change of the setpoint and the signal received on the input G (k), which can be written as:

with

where Ts is a duration over which the rate of change is estimated. In practice, the duration Ts, which is equal to the duration of a sampling interval, is of the order of one microsecond.

Thus, the output u (k) of the correction block 2 can be written as:

The voltage generator block 3, also called "PWM" in the figure, has an input denoted u (k), and two outputs, respectively denoted u (k) and U. The output u (k) is directly connected to the input u (k). In the example shown, the storage battery, denoted "battery", used to generate the voltage U is 12 volts. Also, the output U is equal to 12 times the value of the input u (k).

The estimator block 4, also called "Estimation of G" in the figure, has two inputs, denoted respectively u (k) and I. The input u (k) is connected to the output u (k) of the voltage generator block 3. Input I is connected to output I of actuator block 1.

The output G (k) of the estimator block 4 is equal to the difference between: a signal received from a digital differentiation block, denoted "numerical differentiation" in FIG. 1, which can be written as / (t0), and a multiplication by the gain a of a signal shifted from the input signal u (k), of a register of a temporal unit.

Thus, the output G (k) of the estimator block 4 can be written: G (k) = / (k) - a u (k - 1)

Thus, the value / (k) of the estimator block 4 can be estimated by:

In the example shown, this integral is typically calculated over a time window T of the order of a few dozen times the duration of the sampling period.

As an example, the integral can be calculated by the numerical method known as the trapezium method. The use of this command without a template is justified below. It is based on the possibility of being able to write over a short period of time, the "phenomenological" equation: where:

• v is the derivation order of y, generally equal to 1 or 2, chosen by the operator (it is not the maximum derivation order of y that remains unknown), • the constant parameter a is fixed by the operator so that the numerical values \ au and yM have the same order of magnitude, • G which contains all the structural information on the chosen time frame.

The numerical value of G for an index k can be estimated from the following equation:

where | G (fc) I and [y (v) (fe) I are estimated at the index k, that is, at the instant kTs where Ts is the duration of the sampling period.

The numerical value of ly (v) (& 0Je can be estimated, which can be obtained by measuring the value of output y.These measurements can be affected by noise.

It is thus preferable to use a digital differentiation, which includes noise filtering and derivation.

Figure 2 illustrates the general principle of the implementation of a modelless control which is based on the equation given by the phenomenology model:

In FIG. 2, the instruction is noted y% the command at the index k is denoted ti (fc) and the output, under the effect of the command u (k), of a system, denoted "unmodeled plant" whose model is not known, is noted y. The tracking error y * - is noted there e.

Still in FIG. 2, the control is determined by a block, denoted "Model-free control", from the signal y * and the signal y. More precisely, said block comprises four sub-blocks: a first sub-block, denoted "i-PI", receiving the signals y * and y as well as a signal [G (fc)] e, and generating a signal u ( / c), from the equation:

a second sub-block, denoted "Delay", receiving a signal u (fc) and providing a signal u (k-1), a third sub-block, denoted "Numerical Differentiation", receiving a signal y and generating a signal [ y (v) (fc)] e / from the equation:

a fourth sub-block, denoted "Estimation of G", receiving the signal u (k - 1) and generating the signal [G (/ c)] e, from the equation:

Numerical differentiation: FIG. 3 makes it possible to justify the equation implemented in the third sub-block of FIG.

A linearization at the instant t of the signal allows to write:

A Laplace transformation of this equation gives:

Deriving the equation on the right with respect to the variable s, we obtain:

Hence: From where: where the interval [0, T] is a rather small window of time that is moved to estimate the derivative at each moment.

The trapezoidal method can be used to calculate, in practice, this integral.

This method for determining the derivative has the advantage that it can be used for higher order derivative determinations.

FIG. 4 is an illustration of the advantage of the use of this method with respect to a more conventional derivative estimation method, such as that implementing a first-order low-pass filter followed by a determination of the derivative according to the Euler method.

The graph on the left shows three intensity curves between -1.5 A and 2 A, as a function of time over the interval from 0 to 8 seconds. The graph on the right is a zoom on the interval ranging from 5 to 7 seconds and -0.2 A to 1.6 A left-handed one.

It can be observed that the curve illustrating the implementation of the digital differentiation is on average closer to the perfect curve of the expected derivative, ie of a cosine, than the curve implementing a low-pass filter of first order followed by a calculation of the derivative according to the Euler method.

This curve therefore corresponds to the expected derivative, and it is not necessary to place a filter before the derivation.

FIG. 5 illustrates the devices used to produce the curves of FIG. 4. More precisely, the generation of a noisy signal to be derived, generated from a sinusoid signal to which a white noise signal is added, is observed. .

The three signals shown in FIG. 4 are observed at the output, that is to say a cosine signal (which is the expected derivative of a sinusoid signal), the signal according to the Euler method and the signal indicating implement a numerical differentiation.

The signal according to the Euler method implements a first-order filter, generating a signal filtered from the noisy signal, the filtered signal being then derived by applying the z-transform of the discrete derivative, ie -say

The derivation according to the method presented above uses a period T equal to 30 times the time Ts, that is to say 30 times the duration of the sampling interval.

Choice of the digital differentiation window: FIG. 6 illustrates the influence of the choice of the value of the period T on the determination of the derivative by the method of numerical differentiation.

The graph on the left of FIG. 6 represents an expected derivative, the cosine curve and 3 curves obtained for different values of T, respectively 20 times Ts, 30 times Ts and 40 times Ts.

The graph on the right represents the distances of the three curves obtained at the expected derivative.

It is observed that the greater the value of T, the less the influence of noise on the determination of the derivative. However, the larger the value of T, the longer the delay before getting the expected value.

In practice, a compromise must be found to determine the value of T.

Application to an automatic parking brake ("APB" 1

Figures 7 and 8 study the feasibility of implementing a modelless control for a DC motor.

Figure 7 illustrates that a system whose transfer function has a zero in the positive plane is one that has no minimum phase.

Figure 8 uses the modeling of a DC motor according to the Simulink software to determine the transfer function and check that it has no zero in the positive plane, and therefore that the DC motor is a minimum phase system . It is therefore possible to implement a modelless control for the DC motor.

Simulation: Figure 9 is intended to illustrate the behavior of the model without control compared to a variation over time of a resistance of a system comprising a motor.

The graph on the left shows three curves, respectively a reference intensity, a measured intensity with an erroneous resistance and an intensity measured with an exact resistance, over time from 0.5 seconds to 4 seconds and from -5 A to 35 A We observe that the three curves are very close.

This is confirmed by the graph on the right, which is a zoom of the graph on the left between 0.95 seconds and 1.15 seconds and between 0 A and 40 A.

It is observed that even if the resistance varies from 0.35 ohms to 1 ohms, the behavior of the control process without model is efficient.

As a result, the modelless control process behaves satisfactorily even though the system is evolving.

Figure 10 illustrates the influence of the calculation determination used in the numerical differentiation on the performance of the method, for a curve relating to an electric parking brake.

FIG. 11 comprises two graphs which illustrate the errors obtained between the estimated intensity and the reference intensity for several sets of parameters a, Kp and Ki.

FIG. 12 illustrates an embodiment of a system implementing a control method according to the invention.

The system illustrated in FIG. 12 comprises a DC motor of a brake actuator, capable of receiving a voltage of between 0 and 16 volts coming from an electronic circuit.

The electronic circuit is capable of generating a regulated voltage, from a 16 V battery, and a PWM input from a computer module, programmed for example in Matlab or Simulink.

The electronic calculation module is programmed to implement, when executed by a computer or a processing unit, a part of the control method according to the invention, that is to say as long as a condition of adjustment is true, calculation of a new value u (tl) for the control variable, at a new instant tl, and application of this new input value of the actuator, which value depends on: - the value u (t0 ) of the control variable, - the derivative y (tO) of the output quantity, which is determined from a current value I measured by an intensity sensor module within the actuator, - the rate of variation (y *) (tl) of the setpoint value, between t0 and t1, a factor proportional to the error of tracking of the output quantity, - a factor proportional to the integral of the tracking error of the output quantity, said new value u (t 1) being written in the following form: u (tl) = - + ( y *) (tl) + V (ti;) + Ki fne (t) dt aa Jt0 with G (tl) = y '(tO) - au (tO) where: a, Kv, and Kt are predetermined coefficients, so as to achieve a command without modeling the actuator Proportional-Integral correction type "Intelligent" called "modelless control".

Of course, the invention is not limited to the examples that have just been described and many adjustments can be made to these examples without departing from the scope of the invention. In addition, the various features, shapes, variants and embodiments of the invention may be associated with each other in various combinations to the extent that they are not incompatible or exclusive of each other.

Claims (9)

1. A method of controlling an actuator (1) of a vehicle brake, implemented by a digital computer embedded in said vehicle to produce an application by said brake of a determined clamping force, comprising an iteration of the following steps: - from a value of the desired clamping force, determination (102) of a desired target value y * (tO) (I *) for an output quantity, - application (104) d a control variable of value u (t0) at an instant t0 at the input of the actuator (1), - measurement of a measured value y (t0) (106) of the output quantity, produced within the actuator under the effect of said application, and - as long as an adjustment condition is true, calculating a new value tt (tl) for the control variable, at a new instant tl, and applying this new input value of the actuator (1), which value depends on: • the value u (tO) of the control variable, • the derivative ÿ (t0) of the output quantity, • the rate of change y * (tl) of the setpoint value, between tO and tl, • a factor proportional to the error of follow-up of the output quantity, • a factor proportional to the integral of the tracking error of the output quantity, said new value u (tl) being writeable in the following form: with where: • a, Kp, and K; are predetermined coefficients, • e (t) = y (t) - y * (t) is the tracking error; so as to achieve a command without modeling the actuator Proportional-Integral correction type "Intelligent" called "modelless control". 2. Method according to claim 1, wherein the setpoint y * is determined from a correspondence list comprising a plurality of intensity values each associated with a clamping force value produced by said intensity within said actuator. . 3. Method according to claim 2, characterized in that it comprises, from an initial intensity value chosen according to a clamping force to be applied, an iteration of the following steps: - comparison of said initial value with a database forming the list of matches and containing a series of intensity values associated with a clamping force obtained during a preliminary learning phase, for calculating a modified intensity value according to a coefficient of adjustment sending a motor supply command based on this modified intensity value, according to at least one determined method of motor management, especially in pulse width modulation or in vector control, measurement of at least a parameter of result within the actuator, - estimation of a gain found for said modified intensity value, - according to said observed gain, comparison with said database for calculate a modified value of the adjustment coefficient 4. Method according to claim 2 or 3, further comprising a measurement of a clamping force obtained after application of the control variable u from the setpoint y * and an update of the clamping force associated with said setpoint y * in the correspondence list. 5. Method according to any one of the preceding claims, wherein the set point y * is a current intensity setpoint to be obtained in the brake actuator. 6. The method of claim 5, wherein the value found therein is obtained by measuring the current intensity as generated in the brake actuator. The method according to any one of the preceding claims, wherein the derivative ÿ of the ascertained value is determined therein as an estimate y (t), where: where T is a predetermined parameter. 8. The method of claim 7, wherein the estimate is calculated by discretization of its integral, for example by the trapezoid method. 9. Vehicle or vehicle brake, comprising at least one computer programmed to control one or more actuators according to a method according to any one of the preceding claims.